What problem does it solve? After an epic finishes, no single coding session can see the defects the whole change introduced — god classes grown across nine stories, duplicated logic, spec drift. This Skill reads the evidence an epic left behind (spec, story files, full diff, commits, sprint status, session logs) and produces a sourced findings report plus an honest acceptance verdict. ## Core Features & Use Cases - Evidence-based findings: Every reported finding carries a source reference (file, line, commit, or log); claims without a source are dropped. - Aggregate diff analysis: Derives architecture deltas, duplication maps, god-class growth, pattern divergence, and spec-to-implementation reconciliation across the full epic diff range via deterministic scripts. - Acceptance verdict with gates: Renders accepted, accepted-with-open-items, or rejected against declared or profiled criteria; unfinished stories force a machine verdict of rejected. - Sprint-status integration: Updates sprint-status.yaml atomically with comment preservation, appending owned action items and marking the retro key done, with rollback on validation failure. - Use Case: After an unattended multi-story epic run, ask for a retrospective to get a document listing what was built, which cross-story defects the diff reveals, whether last epic's action items landed, and whether the epic met its own acceptance criteria. ## Quick Start Ask the assistant to run a retrospective on the most recently completed epic, or pass an epic number explicitly for a headless run.
